Anthropic has had a notably active period: the company is working on a multibillion-dollar fundraising round that would value it at roughly $950 billion. That proposed valuation would exceed OpenAI’s March valuation of $854 billion.
Growing enterprise market share
According to a recent report, Anthropic has quadrupled its market share among enterprise customers since May 2025 and has already overtaken OpenAI in that segment. The Claude platform is increasingly favored by corporate clients, in some cases replacing ChatGPT.
Product strategy and Cat (Catherine) Wu’s remarks
Cat (Catherine) Wu, head of product development for Claude Code and Cowork, spoke about the company’s strategy at the Code with Claude conference in San Francisco. Wu joined Anthropic in August 2024 and has reportedly played a central role in evolving Claude from a simple chatbot into a developer tool and a more versatile product.
Wu said the company deliberately does not obsess over competitors when shaping product strategy. In her view, constantly tracking rivals risks falling behind; Anthropic therefore focuses on staying at the forefront of AI development.
Rapid model releases and changing rollout approaches
Anthropic released at least six models last year and is maintaining a similar pace in 2026. Wu noted that model performance continues to improve steadily, but the way releases are handled may change: not every new model will be broadly exposed immediately.
As an example she cited the Glasswing initiative. Launched in April, the program made Anthropic’s new cybersecurity model, Mythos, available only to a narrow set of partners. That select group includes Amazon, Apple, CrowdStrike and Microsoft. The company says models intended to identify vulnerabilities in codebases pose too great a risk to be widely accessible.
The role of agents and challenges of management
Wu has previously argued that the future of work lies with AI agents, with employees managing fleets of such agents. She tempered that view in her recent remarks, emphasizing that managing agents is difficult when users lack domain expertise. Understanding and correcting an agent’s error is similar to managing people — you need to determine whether the agent misunderstood an instruction or whether the prompt was insufficiently precise.
Asked whether agent adoption will reduce headcounts, Wu provided a reassuring perspective: she believes agents will mainly take over monotonous, repetitive tasks (for example, responding to routine emails), freeing humans for substantive and creative work.
Proactivity becomes the next focus
Wu sees proactivity as the most important near-term development. While synchronous development dominated last year, 2026 will emphasize automated routine processes, such as handling customer-support tickets. The next major step will be for Claude to autonomously recognize what a user is working on and then configure the necessary automations on its own.
Overall, Anthropic’s approach combines rapid, continuous model releases, stronger positioning toward enterprise customers, and cautious, partner-limited rollouts for sensitive capabilities — while moving toward proactively acting AI agents.
Key facts and dates
- Planned fundraising: multibillion-dollar round; target valuation roughly $950 billion.
- OpenAI March valuation: $854 billion.
- Cat (Catherine) Wu joined Anthropic: August 2024.
- Glasswing launch: April 2026; Mythos test partners: Amazon, Apple, CrowdStrike, Microsoft.
- Anthropic released at least six models last year.
Tags: cybersecurity, artificial intelligence, fundraising, OpenAI, developer, chatbot, investment, tech companies, AI agent, Anthropic
